Let’s go back to industrial numerical simulation to conclude this first volume. It has been used regularly – if not intensively – for only a few decades. Among the engineers who retired around 2010–2015, some of them made their first steps with this technique and themselves programmed and used the first calculating machines, working with punch cards! A technique closer to the analytical engine of Charles Babbage and Ada Lovelace than the supercomputers used nowadays in HPC. Today, the computing power and ease of use of simulation codes make numerical simulation accessible on a laptop computer! The mathematical models with a few dozen unknowns used more than 30 years ago to dimension real structures nowadays serve as case studies for the practical work of engineering students [SIG 15], while industrial models now reach several hundred million unknowns… In a few decades, engineers have built rules of good practice for this technique [DUB 16] and they use digital simulation without knowing in detail the algorithmic subtleties, which are the property and exclusivity of the tool editors. To the best of their ability, engineers practice it with a critical eye on the results produced by a calculation.

One of the most important issues concerning digital simulation today remains its practical use and possible evolutions in relation to data sciences. As we have mentioned several times in this first volume, numerical simulation is not intended to replace humans, but to assist them in a complementary way in the practice of their professions and in the application of their knowledge. Machines and algorithms contribute to simulation with computing power and efficiency that is inaccessible to the human brain, whose faculties are mobilized for many tasks. The criteria for analyzing the results of a simulation are based on human decisions, their experience and the consent of engineers.

Calculation error – and misinterpretation of a calculation – by engineers using numerical simulation is possible. Like errors in design, diagnosis, judgment or investment, they are also human – and they also accompany the scientific process [FIR 15]. To date, simulation softwares do not decide for engineers. Machines are not yet replacing humans in the design of complex systems – will they ever be able to do so, and how? Engineers have given themselves the scientific means to build the framework for this use, driven by construction requirements: safety, reliability and profitability required by the industrial sector… These requirements do not totally contradict the ethical conception that engineers have of their profession and its aims. In order to support the evolution of this technique, it remains relevant that:

• – engineering students receive training in numerical modeling, allowing them to experiment with the hypotheses of a calculation;
• – engineers maintain their critical judgment and physical sense when interpreting calculations;
• – simulation experts stay in constant dialogue with the designers who use the results;
• – calculation practitioners are involved in innovations in this field;
• – citizens and their political representatives are taught about the innovations and applications it allows.

The practice of numerical simulation for engineering, in its broadest sense, is limited. We have already highlighted some of its limitations. Let us recall the main ones.

A right model… or just a model? A model, based on equations supplemented by data, is a representation of valid reality based on known and accepted assumptions. The physical characteristics necessary for some models are obtained under specific experimental conditions and their use in other configurations is an accepted practice, with knowledge of the approximation. Not all objects in the real world are equally suitable for computer modeling. This is essentially the result of choices, based above all on the practice and physical sense of simulation practitioners.

The validation of certain calculation methods is sometimes carried out blindly. On a simple example, several teams or calculation code editors propose their simulation results. These are then compared with reference results. This exercise pushes the tools to their limits and also helps build their reliability. Let us entrust a modeling of a complex mechanical system to two engineers using the same calculation tool. Depending on their experience with the simulation, the tool, their understanding of the problem and the modeling assumptions they retain, they may find different results… and their interpretation may not be identical!

The calculation produces a result, always containing an error that engineers consider acceptable for the purpose for which the calculation is intended. In this sense, the error is always related to a practice and a measurement scale. Thus, calculating the vibration of a simple shaped part made of an easily characterizable material is carried out with an accuracy of a few tenths of a percent of Hertz. Estimating the noise level of a marine platform under the effect of vibrations induced by current turbulence is best obtained with a few percent of decibel – the scale is in this case logarithmic, which means that the difference between the calculation and the modeled situation is numerically larger! In both cases, it is what engineers expect from simulations that matters: they can be considered reliable, as a practice shared by a community and validated by experience. This in no way prevents us from trying to push their limits, when it is technically or economically necessary – and possible…

No calculation method to date can claim a form of universality and solve the quadrature of the calculation: to simulate quickly, with great precision, a complex object or phenomenon, by using the cheapest possible computer means. The methods developed by the researchers are based on a panel of algorithmic solutions and engineers have a set of digital tools that they use according to the problems they face.

Data: a key resource in the 21st Century for simulation (as for all economic activities). Whether it is industrial or concerns other fields mentioned in this book, the quality of a simulation and the predictions it authorizes depends in part on the quality of the data required to build the models.

From medical confidentiality to military confidentiality, including the protection of human freedoms and industrial knowledge, data are sensitive. They become an important asset for any organization wishing to learn from them. Communities, businesses and citizens are becoming aware of their ethical importance, as well as their economic value. They are thus sources of greed: some of the computer attacks today focus on the weaknesses of data protection and archiving systems. Tim Morris, director of NAFEMS, says:

Data are very strategic and economic in nature. This is still what makes many industrial groups reluctant to use computing resources such as ‘cloud computing’, which could benefit some simulations…

In companies, the data useful for numerical simulation is very varied: test results, calculations and a whole corpus of documents and regulations (such as design notes, calculations, architecture plans, test reports, control images, etc.). Computer-assisted learning could reveal interesting trends for more optimal design, improved risk or resource management.

However, it is not enough to have reliable data, where they exist, and data analysts to produce a robust model – they must also be used in a relevant way! According to Philippe Bonnelle, head of a modeling department at Total:

The functioning of an industrial entity as complex as an oil refinery involves ‘multi-physical’ phenomena. In order to simulate its operation, it is necessary to have analysts interact with specialists in these installations. Their multiple skills (in automation, chemistry, process engineering, etc.) are essential to give meaning to data.

Digital simulation is carried out by humans, in human organizations. Implementing a digital simulation technique requires skills at different levels. The way women and men look at simulation in companies depends on their knowledge, personality, understanding of techniques, development needs… and, of course, the resources allocated. Tim Morris notes that the organization of the calculation process is as important as technical competence. Each entity develops its own culture and pace of appropriation of the technology and its future changes:

Computer simulation is developing in almost all stages of design or industrial production. Aware of the risks of entrusting computers alone with the ‘responsibility’ of dimensioning the most complex mechanical systems, almost all the major industrial groups using simulation accompany their practice with certification and quality procedures for calculation procedures that are not limited to software alone. The training of engineers in scientific computing and its technological breakthroughs, their involvement in the development of new ways of calculating, is as important as the efficiency of digital and computer platforms. This approach is gradually being imposed on all users of the simulation…

New measurement and memory techniques help to record and store sensitive data with great reliability. A range of data analysis tools is available and interface with numerical simulation. The emergence of new techniques is changing practices that have sometimes been established for a long time. Human beings remain at the heart of these changes, as François Bodin, researcher at INRIA, reminds us:

Communities, companies, laboratories: organizations facing the limits of certain practices will develop their scientific computing solutions, based on a hybrid use of digital techniques. These techniques are mastered by scientific communities that are sometimes compartmentalized: future innovations will require going beyond organizational limits!

The digital transition is also a question of generation and choice. Like any change, it is desirable that it be accompanied – if possible at human tempo – to be accepted in the best way. Some of the limitations mentioned above may be pushed back in the future, with solutions that researchers and engineers can imagine… before testing their actual effectiveness. How will these improvements be implemented and how will numerical simulation – and the techniques it helps to develop – be used?

The power of algorithms contributes to the development of digital techniques which, as they become widespread in all areas of human life, raise the question of their replacement by machines [CAR 14] or, more essentially, of their freedom [SAD 18]. Part of this technical progress concerns everyone when it is not accompanied by true social, ecological or ethical progress. It is rightly associated with job losses – and it is difficult to say today that the digital transition will be accompanied by so many creations – and it is accompanied by irreversible environmental destructions. They call for a rethinking of the organization of work and the place of humans in a transformed society and environment; as well as the place that the latter gives to those for whom adaptation to this new world can present difficulties. Some trades, based on repetitive tasks that make work more painful than fulfilling, are directly affected by the automation of the economy. Among the lowest paid, their vulnerability to technological innovations subjects those who occupy them to a double penalty, that of exploitation and insignificance, thinks, for example, Yuval Noah Harari [HAR 18a]. Other professions, presented as less vulnerable to technological upheavals, are likely to evolve with algorithms increasingly assisting humans in their activity, sometimes to the point of being more efficient than they are, and depriving them of their freedom to act or to contribute to the progress of the world by being deprived of the meaning they wish to give it. Industrial and service workers, truck and taxi drivers, farmers, analysts in the financial, insurance and health sectors: many trades are becoming vulnerable to advances in automation and AI. It should be noted that the complexity and precision of movement allowed by the hand and the adaptability of the brain and body required in many trades still distinguish humans from machines: “the profession of the plumber (or carpenter) is not about to be replaced by a robot!” [GEL 18]. The complementarity between human and digital skills makes it possible to think about future changes more broadly: “the labor market in 2050 could thus be characterized by human-IA cooperation rather than competition” [HAR 18a].

While it is still difficult to quantify the impact of technical disruptions, trying to anticipate their effects and proposing solutions becomes a necessity [CHE 18, DUR 17, LAU 17, PIR 16]. The balance between job creation and job destruction is at the heart of the digital transition. Its support is above all a question of political choices: the adaptation required by the hectic pace of technological innovations cannot be satisfactorily achieved for humans without the latter having real social stability, which is the result of decisions, as well as struggles, that we can hope for collectively. Despite all these questions, we can also imagine that there is no shortage of work at the beginning of the 21st Century, as humanity as a whole is facing many challenges. While there is no shortage of work for humanity, money to pay is said to be scarce. It is more likely to be kept away from humans whose contributions to its profitability are considered the least effective and from projects that are among the most useful for the survival of humanity – projects that are inherently less remunerative in the short term. In 1891, the French writer Émile Zola (1840–1902) published Money, the penultimate volume of his literary fresco telling the story of a French family, the Rougon-Macquart’s [ZOL 91]. A spirit of innovation or manipulation, an entrepreneur against a speculator, power and communication games: money has no smell and is the source of many people’s happiness! It allows visionaries to realize their dreams, some of which are beneficial to humanity, and it also locks its harpoons, and others, in a nightmare, each giving it meaning through its actions. For Zola, money is above all a means, not an end, and the economy is not governed by natural laws, but results from political choices. Physics reveals mathematical models explaining how realities are imposed on us. We have no choice but to accept them and adapt to them. We can escape it theoretically through an experience of thought – a simulation – or concretely with the help of technology. A plane will free us, for a time, from the contingency of gravity, with a minimal risk of accident. However, the physical world constantly calls engineers’ dreams to reality. The objects they design start from the constraints of reality and pass the test of the operational – we have mentioned and demonstrated this on several occasions in this book. When engineering projects, sometimes extremely costly, come up against the realities of technical impasses, insufficient economic resources or financial profitability, it is up to the industrialists and the States that initiated them to learn from them… which they sometimes do [CHE 09]. Are we really doing the same with economic theories and financial practices? Do we easily renounce excessive political orientations when social and environmental realities show that they lead to more instability – and pose a risk to our future? Do we question the financial logic that requires constant and high return on invested capital, when we live in an essentially dynamic, interconnected and limited world? The international competition of social systems needed for unlimited growth is supported by ultra-liberal policies that impact all work organizations and obstructs many entrepreneurial opportunities – to develop a long-term vision and to provide the means to collectively achieve it in order to meet the challenges facing humanity.

The latter are polymorphic, mixing political, ecological, technical and ethical issues. Let us retain here some of those who are committed to its survival.

Maintaining access to essential resources, satisfying the vital needs of a human population of more than 7 billion people, nearly 11 billion by the end of the century, in a world whose material wealth, whose waste and over-consumption we are becoming aware of, has never been so important – and somehow poorly distributed [PIK 13]. Progress in agriculture has contributed to better nutrition for all humans, despite significant population growth. Although it still faces problems of undernourishment (Figure C.1), humanity in the 21st Century no longer faces famines due to natural phenomena (droughts, floods, crop predators, etc.). Such episodes occur due to human decisions (armed conflicts, organization of the food production and distribution system, etc.) [PIN 18b]. Feeding humanity in 2050, when it has 9 billion people, is not totally beyond our reach… To date, however, we are living by consuming resources at a higher rate than our planet can provide: this is the observation made each year by the NGO Global Footprint Network, which has helped to popularize the concept of “Earth Overshoot Day”. Taking on a new dimension in the age of data science, it was developed following the concept of the “ecological footprint”, proposed in the 1990s by Swiss researcher Mathis Wackernagel [WAR 17].

COMMENT ON FIGURE C.1.– A significant proportion of humanity in the poorest countries continues to suffer from undernourishment. By 2050, the world population is projected to reach nearly 9 billion people. With an appropriate production and distribution system, it is quite possible to satisfy the food needs of all humanity, by limiting the impact of the most environmentally aggressive agricultural practices. This requires imagining new production techniques and assessing their consequences, making investments and supporting their implementation – and launching ambitious policies to fight poverty (Source: http://www.fao.org).

The calculation consists in assessing the planet’s organic production capacity, that is its capacity to provide natural resources, on the one hand, and human consumption, on the other. Earth Overshoot Day symbolizes the date from which humanity exhausts its annual natural resources each calendar year… In 1990, it was estimated to be December 7 and in 2018, August 1… almost four months earlier! The decline in this date is constant and indicates that meeting human needs requires more than the Earth can provide (Figure C.2). At the global level, all countries have an unequal contribution to this resource consumption. Modeling based on equations or data helps to identify the state of resources – and to assess agricultural policies and practices, for example.

COMMENT ON FIGURE C.2.– According to simulations by the NGO Global Footprint Network, meeting human needs from the planet’s biological resources required 1.7 Earths in 2018, compared to 1 Earth in 1960. This global indicator makes it possible to become aware of certain consequences of our lifestyles – and their unequal influence. Overall, if humanity lived at the pace of the richest countries, it would consume the resources of 1.8 to 5 planet Earths. In recent years, there has been a slowdown in the progression of Earth Overshoot Day, a sign that our consumption habits are evolving towards a more sustainable model? (Source: https://www.overshootday.org/)

Anticipating and combating the effects (and causes?) of ecological and climate change that can very quickly upset major balances is, for some, one of the major challenges of this century [HAR 18a]. In order to preserve the environment in which humans live and to propose solutions for action, it is necessary to identify the proportion of these changes attributable to human activities. Adapting lifestyles to this change, in other words saving material resources and adjusting their use, is one of the levers to be activated.

Human activities are responsible for greenhouse gas emissions, including methane and carbon dioxide. Since the industrial era, they have been growing at an exponential rate (Figure C.3)!

COMMENT ON FIGURE C.3.– The figure represents CO₂ emissions over an extended period, from the second half of the 17th Century to the present, according to various contributors. In 2015, China, followed by the United States and the EU-28 (countries that were member states of the European Union as of 1st July, 2013), together represent more than a third of the world’s population and account for more than half of CO₂ emissions. In nearly 115 years, these have increased twentyfold, from 2 billion tons in 1900 to 36 billion tons in 2015. The most recent data (2014–2016) suggest that annual CO₂ emissions are stabilizing. However, this slowdown is too recent to determine whether it is a peak, possibly followed by a reduction, or a plateau. A significant part of this stabilization is attributable to China, whose emissions are stabilizing. More specifically, emissions from the richest countries in Europe and America peaked in the early 2000s and the question is now to assess the current and future trends (declining or rising, and at what pace?). In 2015, CO₂ emissions were distributed globally: 60% in terms of energy production, 15% transport, 10% housing and commerce, 8% agriculture and 7% industry (Source: Our World in Data, https://ourworldindata.org/co2-and-other-greenhouse-gas-emissions).

The increase in their concentration in the atmosphere contributes to rapid climate change, which can be quantified through simulations. The rate of observed warming and the consequences it has on all human activities are at the heart of the current questions of those – political leaders, economic, intellectual and scientific leaders as well as ordinary citizens – who are thinking about the worlds of tomorrow. The search for solutions to reduce greenhouse gas emissions is one of the challenges common to many sectors (energy, agriculture, transport, etc.) and numerical simulation is one of the techniques that can be used to achieve this objective.

There is a goal to ensure the best possible living conditions for humanity and give it hope for its future, where some have been predicting in recent years that it will be bleak because of ecological issues. While some dream of giving humanity super powers allowed by technology, the latter primarily allows for the development of solutions that help to reduce some of the most common causes of death (Figure C.4).

COMMENT ON FIGURE C.4.– The causes of mortality and the standards of living and wealth of their populations vary across countries. Globally, the majority of deaths, more than 70%, are due to non-communicable diseases, primarily cardiovascular or respiratory diseases, cancers and diabetes. They are the most common causes of death in the richest countries – whereas infectious diseases or malnutrition are more common in the poorest countries. Beyond their strong emotional impact, armed conflicts, terrorism and natural hazards are fortunately only marginal causes of death: humanity as a whole thus lives in the most peaceful and secure world it has ever known [PIN 18b].

Some digital techniques will eventually provide models and valuable data on humans and their health. Accompanying and benefiting from the development of “biotech” and “infotech”, they promise new opportunities for humans. Their possible generalization also raises new ethical and political questions [HAR 18a, PIN 18b]. In the second volume of this book, we show how digital simulation contributes to science and technology in order to provide answers to some of these challenges: feeding, caring and understanding humans, producing the machines and energy they need, and studying the Earth, oceans, climate and the Universe.

We opened this first volume with Jules Verne’s enthusiastic look at technological progress and our work wishes in some respects to illustrate it, without however avoiding the many questions raised by the use of techniques that numerical simulation help, among other things, to develop. Our work is, of course and largely, incomplete, the purpose here being to talk about a brick that fits into a larger technical building. It is up to each individual to complement it with other resources and other perspectives in order to appropriate its complexity and, perhaps, contribute to the choices of society.